PROJECT SUMMARY: This research uses cigarette smoking during pregnancy (SDP) for investigating risk- processes in addiction. While population-level trends on sociodemographic risk factors for SDP and the adverse effects on offspring associated with maternal SDP are well established, the more complex task of understanding individual risk-prediction and its clinical application remains incomplete. I will implement a ?big data? approach using cross-linked research data (from a large female twin cohort, followed prospectively from median age 15; new data collection with Native American mothers) and state individual-level vital and driving records data (SDP, sociodemographics, geocoordinates) to characterize the interplay between individual and environmental factors conferring risk or protection for continued smoking throughout pregnancy. AIMS: (1.1) characterize interplay of sociodemographic and neighborhood influences on SDP using state data on 100,000s of births, and examine potential confounders of racial disparities in SDP; (1.2) use cross-linked research and state data to characterize effects of individual risk-factors (nicotine dependence, psychiatric and trauma history) as predictors of SDP, including their interplay with a summary neighborhood/sociodemographic risk-score derived from in (1.1); and cross-validate birth record SDP report; and (2) ascertain from state vital records a new cohort of American Indian (AI) mothers (a cohort with very high rates of SDP) for a pilot retrospective study, to extend findings in 1.2 to an understudied group and to address limitations of existing research?s failure to identify specific cultural factors associated with SDP in this group. METHODS: State birth record/driver?s license data (SDP, teen births, DUIs), aggregated to the census tract, will be used for prediction of SDP risk, separately for mother?s residence during teen years and at time of pregnancy, to supplement standard Census/American Community Survey socioeconomic disadvantage predictors. Logistic regression will be used in (1.1) to generate a summary risk score for analyses with research data in (1.2), reducing the risk of false-positive findings. Standard assessments optimized for use with AI populations will be used in retrospective interviews of a small series of AI women. PREDICTED RESULTS: Demonstration of race-specific interactive contributions to SDP risk of individual, sociocultural, sociodemographic, and neighborhood characteristics, allowing tailored individual risk prediction. CANDIDATE, TRAINING: The applicant is an MD-PhD candidate at a leading institution for training successful independent physician-scientists, working with a strongly committed mentoring team, and using unique resources. The proposed training plan provides new conceptual and technical training with an outlined set of career development activities, workshops, and formal didactic opportunities, to support development as an independent physician-scientist working to improve understanding, assessment, and treatment of addiction, particularly in understudied/underserved groups.